Abstract
Across six experiments, we studied the relationship between target-distractor discriminability and search efficiency, when the only differentiating feature between objects is color. We assessed target-distractor discriminability by measuring the distance in CIELab space between nearly equiluminant target and distractor stimuli, over a range where the distance in CIELab space measures the perceptual discriminability between two colors. We presented stimuli at either two or three different eccentricities, and we cortically magnified the stimuli to equate visual processing times. Finally, to prevent serial search, we selected color distances that were sufficiently large such that peripheral vision could accurately detect those color differences. Results from Experiment 1 demonstrated that there is an inverse relationship between logarithmic search efficiency and target-distractor color distance, as was proposed in Lleras et al. (2020). This relationship was confirmed in four additional experiments. Experiments 2-6 explored whether different stimuli conditions could break or modulate that relationship. Experiment 2 found no evidence that presenting distractor colors only to one side of the target’s color on the color circle would improve search performance, as proposed by optimal tuning theories of feature-based attention and by relational tuning accounts. Performance was identical to when distractors were picked from both sides of the target color. In Experiments 3 and 6, blocking distractor colors resulted in a 50% improvement in search efficiency across the entire range of colors. In Experiment 4, target and distractor colors varied randomly from trial to trial. Search performance was no longer logarithmically related to set size and RTs were unusually elevated. Experiment 5 studied the improvements in search efficiency when the same color is repeated over the entire experiment. Across all experiments, we found no evidence that feature-based attention to color had altered the appearance of the stimuli, or that features had been boosted or suppressed to improve performance.